Medulloblastoma is the most common pediatric malignant brain tumor. Although the 5-year survival rate has improved to 50~80% with neurosurgery, chemo- and/or radio-therapies, such non-specific treatments often lead to severe side effects on brain development in young patients. Targeted treatment methods are in urgent need to selectively block tumor growth without affecting normal brain development. While much research focuses on understanding molecular aberrations in tumor cells, another worthy focus is deciphering intricate interactions between tumor and neighboring cells. These interactions include a crosstalk among different cells in order to ensure proper form and function during development and adulthood. In parallel to normal organ development and function, the crosstalk between tumor cells and their neighbors is critical for sustaining tumor growth. However, research in this area is still under-represented, mainly due to difficulties in unambiguously identifying tumor versus neighboring cells in a complex tumor mass. To overcome such hurdles, the proposed studies use a mouse genetic system that can mark all tumor cells with green fluorescent protein (GFP), thereby distinguishing them from colorless neighboring cells, allowing dynamic analysis of green mutant cells over time. Medulloblastoma has been shown to originate from granule neuron precursors (GNPs) based on their surface location, cellular markers, and transcriptional profiling. Since GNPs are known to only give rise to granule neurons, we were surprised to find that labeled tumor cells in our mouse model were not homogeneous. There were GFP-labeled glial cells in addition to GNP-like tumor cells in the tumor mass. To tease apart the functional relevance of these glial cells, we used genetic methods to specifically ablate them without harming GNP-like tumor cells. Amazingly tumors disappeared within a few days after glial ablation. Taken together, we hypothesize a community building behavior in medulloblastoma: mutant GNPs broaden their developmental potentials to give rise to glial cells, which in turn provide critical support to tumor GNPs for them to survive and thrive. The proposed studies aim to elucidate both the spatio-temporal dynamics of the tumor regression phenomenon as well as investigating the molecular crosstalk between tumor glia and GNPs, which appears essential for tumor growth. The blockade of such crosstalk should allow one to devise effective therapeutic methods to cut off support to the tumor via supporting glial cells. These studies should not only provide critical basic knowledge for the community building phenomenon in cancer biology, but also facilitate the design of highly specific and effective intervention methods that aim at disrupting such behaviors in medulloblastoma.